Where the Old Web Went: Link Rot and the Quiet Collapse of Internet History
Click an old link — a citation in a 2005 blog post, a reference in a university essay, a bookmark you saved a decade ago — and there is a decent chance it will lead nowhere. A 404 page. A parked domain. A generic hosting placeholder. Sometimes an ad-choked squatter site that has nothing to do with what used to live there.
Researchers who have followed hundreds of thousands of links across the web keep arriving at the same uncomfortable conclusion: the internet forgets faster than we think. The phenomenon has a name — link rot — and it is quietly gutting the historical record of the first three decades of the online era.
What link rot actually is
Link rot is the slow decay of hyperlinks as the pages they point to move, change or disappear. It has two close cousins. Content drift is when a URL still works but the page behind it has changed so much that the original reference no longer makes sense — a news article rewritten, a Wikipedia section deleted, a company "About" page rebranded. Reference rot is the combination of the two: the link either dies or the meaning behind it does.
Individually, a broken link is a small inconvenience. Collectively, they represent something bigger — the erosion of the connective tissue that made the web a web in the first place. Every dead link is a small hole in the map.
Why old websites disappear
Websites do not vanish for one reason. They vanish for dozens, and most of them are prosaic rather than dramatic.
- Hosting stops getting paid. A personal site on a small hosting plan quietly lapses when the credit card expires, someone changes jobs, or an enthusiast loses interest. There is often no one left to notice.
- Domains expire and get squatted. A lapsed .com or .com.au domain is quickly snapped up by speculators. The old content is gone; the URL now points at parking pages or SEO spam.
- Platforms shut down. When Yahoo closed GeoCities in 2009, it wiped out one of the largest concentrations of amateur creativity the web had ever produced. Similar closures have taken down MySpace music uploads, Google+ posts, Vine videos, countless forum communities and, more recently, huge swathes of Tumblr and early blog platforms.
- Redesigns break URLs. Every time a news outlet, university or government department overhauls its site, thousands of old article URLs can silently 404 because the new CMS uses a different link structure and nobody bothered with redirects.
- Content moderation and legal removal. Copyright complaints, defamation actions, GDPR-style "right to be forgotten" requests and platform policy changes all pull material offline permanently.
- Companies die or get acquired. Startups fold, brands are absorbed, and the acquirer rarely cares about preserving the old blog, help centre or user forum.
None of this is malicious. It is the natural entropy of a medium built on rented infrastructure. Paper libraries survive because someone owns the building and pays the roof bill for a century. Websites survive only as long as somebody keeps renewing a domain and paying a bill they may not even remember setting up.
Why this matters more than it sounds
It is tempting to shrug. Old GeoCities pages about anime and glittery <marquee> tags are not the Library of Alexandria. But link rot has serious downstream effects.
Academic research is one of the clearest casualties. Journal articles increasingly cite web sources, and studies have repeatedly shown that a significant share of those citations become unreachable within a few years of publication. When a footnote points nowhere, the evidence base for a claim quietly evaporates.
Journalism is another. News reporting increasingly builds on earlier reporting, embedded tweets, official statements and government pages. When any of those disappear, the paper trail behind a story disappears with them. Court cases and public inquiries have been complicated by exactly this problem.
There is a cultural dimension too. The early web — personal homepages, fan sites, hobbyist forums, independent zines — captured how ordinary people thought, argued, joked and built community online. That material is disproportionately vulnerable because it was made by individuals rather than institutions. When it goes, we lose the social history of the internet's first generation.
Who is trying to save the web
The best-known rescue effort is the Internet Archive's Wayback Machine, which has been crawling and snapshotting web pages since 1996. Type in an old URL, pick a date, and you can often recover a version of the page as it looked at a particular moment in time. It is not comprehensive — the Archive cannot crawl everything, and it honours removal requests — but it is the closest thing the web has to a working memory.
Australia has its own quietly impressive contribution. The National Library of Australia runs Trove's Australian Web Archive, descended from the PANDORA project it launched in the mid-1990s. It preserves Australian government sites, election material, community pages and cultural content that would otherwise be lost when funding, interest or hosting runs out. State libraries and universities contribute similar collections.
Other efforts sit at different scales. Archive Team, a loose collective of volunteers, races to scrape sites when a shutdown is announced. Tools like Perma.cc let researchers and courts create permanent, timestamped snapshots of the pages they cite. Some publishers automatically archive every outbound link in their articles.
None of this is enough on its own. Web archiving is expensive, legally messy across jurisdictions, and technically hard — modern sites are less like documents and more like software, rendered on the fly from databases, APIs and JavaScript that archival crawlers struggle to capture faithfully.
What we can do individually
You do not have to run an archive to help slow the rot.
- When you cite a web page in anything that matters — an essay, an article, a report — save a Wayback Machine snapshot at the same time and link to both.
- If you run a website, take redirects seriously. When you change your URL structure, map the old paths to the new ones. Cool URIs, as the old W3C mantra put it, don't change.
- If you have old blogs, forums or personal sites, consider exporting them and depositing copies with a library or archive rather than simply letting the hosting lapse.
- Support the institutions doing this work. The Internet Archive is a non-profit; the National Library relies on public funding.
A web with a memory
The web was originally sold as a permanent record — everything, forever, at your fingertips. What we have instead is closer to a vast, restless conversation, most of which is never written down twice. Sites vanish faster than we can index them, and the earliest generations of internet culture are already largely gone from the live web.
Following the trail of dead links is less a technical exercise than an act of remembering. Each 404 is a small argument for taking digital preservation seriously — not as nostalgia, but as basic infrastructure for a society that now conducts most of its public life online. The old web went where anything unmaintained eventually goes. The question is how much of the current web we are willing to let follow it.